Process Enhancement of a Cr (VI) Remediation Method to Minimize the Hazardous By-products in the Treated Water

(1)

Process Enhancement of a Cr (VI)

Remediation Method to Minimize the

Hazardous By-products in the Treated

Water

AB Masombuka

Dissertation submitted in fulfilment of the requirements for

the degree Masters in Environmental Science with

Hydrology and Geohydrology

at the North West University

Supervisor: Mr. Nicolaus van Zweel

Co-supervisor: Mr. D van Schalkwyk

Examination November 2018

29883903

(2)

I declare that this research project is my own, independent work. This research report is being submitted in partial fulfilment of the requirements for the degree, Master of Science (Research) in Hydrogeology at the University of North West, Potchefstroom. It has not been submitted for degree awarding or examination purposes at any other university.

(Signature)

Abel Bongane Masombuka; B.Sc. (Hons), Cand. Sci. Nat. (No 115797)

(3)

ACKNOWLEDGEMENTS

I would like to thank Mr Morné Burger, Mr Collen Nkosi, Dr Altus Huisamen and the management of Geo Pollution Technologies for allowing me to undertake this study and for their continued support throughout the project. Mr Morné Burger conceptualised and designed the remediation system and I congratulate him on his innovative thinking and sound scientific knowledge and expertise. Mr Collen Nkosi conducted field work and captured the data accurately and his background of the study area helped immensely. Dr Altus Huisamen significantly contributed in building a conceptual site model, transport model and numerical flow model and reviewing the project report prior to submission to the University; his contribution is highly appreciated. I thank my colleagues, Mr Christo Gouws and Ms Vevanya Naidoo for assistance with the fieldwork and continued support. I would like to thank my partner, Sibongiseni Masombuka and my family for their support during my time of research. A special word of thanks to my supervisor Mr Nicolaus van Zweel for enormous inputs and direction on this project.

(4)

ABSTRACT

A Cr(VI) contamination at the study area is related to the leaking baghouse slimes dam which was decommissioned 18 years ago, in order to manage and control excessive Cr(VI) contamination impacting the aquifer, and thereafter impacting the off-site receptors. The baghouse slimes dam was decommissioned in the year 2000 and covered with polyethylene thermoplastic to prevent leaching prompted by rain water and to prevent Cr oxidation. Cr(VI) contaminated groundwater is pumped from the underlying aquifer system to the two existing surface treatment settling ponds and thereafter the water is treated by means of dosing system (FeSO4) and electrochemical or reducing system. These treatment systems are

effective in managing or controlling the Cr(VI) contamination. However, neither system can work effectively independently. Hence, the installed liner is another mechanism that was implemented to enhance the remediation systems at the study area.

The main purpose of this project is to determine the effect of the polyethylene thermoplastic liner on the seepage water quality observed in the monitoring boreholes. The plume movement from the source area to the impact monitoring boreholes was modelled in order to ascertain the plume mass, plume movement during abstraction and the plume capture zone. The water chemistry data was used to achieve the objective of the study.

The definite decreasing Cr(VI) concentrations in nine (9) of seventeen (17) existing monitoring boreholes were observed from the year 2000 to 2017. This confirms that the installed liner was capable of ceasing the recharge in the slimes dam and minimizing the chromium oxidation. Six (6) of the seventeen (17) monitoring boreholes showed fluctuating concentrations, which could be attributed to interaction of groundwater chemistry and geological formation. The remaining two (2) monitoring boreholes showed an increasing trend, that stipulates that the plume is localised within that area and the Cr(VI) concentrations trend in these boreholes indicate a potential gradual decrease. Based on the Mann-Kendall trend analyses result, the installed liner is proved to be effective on the seepage water quality.

Given that the plume has reached the impact boreholes, it was deemed necessary to ascertain plume movement, plume mass and the plume capture zone via transport and numerical model using Groundwater Modelling System (GMS10.0).

(5)

taking place at the study area and they were within an acceptable error. Sensitivity analysis indicated certainty for most stressed parameters. Only two off-set values were evident for recharge and hydraulic conductivity of layer number two.

Based on the modelling results, the movement of the plume is controlled by sorption process. The plume was characterised by means of considering and disregarding sorption. This exercise proved that the plume is moving rapidly when sorption is not considered. The plume mass was also calculated with and without sorption and a difference in mass was observed. When sorption is considered the mass over time is larger than when it is not considered. This proved that when sorption is not considered in the numerical or transport model, inaccurate predictions of plume mass and movement can be calculated and this can lead to negative impacts on plans and finances of the mine.

(6)

LIST OF FIGURES

Page

Figure 1: A flow diagram from Riekkola-Vanhanen indicating the most common process

steps utilised for FeCr production in SA (Beukes et al, 2010). ... 5

Figure 2: The oxidation state and potential energy of chromium ... 7

Figure 3: The relationship between Eh and pH ... 8

Figure 4: Three dimensional model illustrating contaminant transport by advection (Strassberg, et al., 2011) ... 9

Figure 5: Mechanical dispersion transport on the microscopic scale (Amy Allwright, 2014) ... 10

Figure 6: Hydrodynamic dispersion transport on the microscopic scale (Amy Allwright, 2014) ... 10

Figure 7: Site location ... 13

Figure 8: Overview of the vadose zone and geology of the area ... 15

Figure 9: Mean monthly rainfall and evaporation data as recorded at the Hartebeespoort Dam Meteorological Station (A2E001) over a 17-year period (Department of Water and Sanitation, 2017) ... 20

Figure 10: Local geology of the study area (L.N.J. Engelbrecht & M.BL Direkteur, Geological Survey)... 22

Figure 11: Groundwater flow direction of the study area ... 23

Figure 12: Site showing monitoring points in relation to the Old baghouse slimes dam ... 25

Figure 13: Conceptual Site Model (without polyethylene thermoplastic liner) ... 28

Figure 14: Conceptual Site Model (with polyethylene thermoplastic liner) ... 29

Figure 15: Vertical delineation of the study area ... 31

Figure 16: Lateral delineation of the study area ... 34

Figure 17: Concentrations of Cr(VI) across the study area (2015) ... 40

Figure 18: Concentrations of Cr(VI) across the study area (2017) ... 41

Figure 19: Illustration of cation and anion compositions of all the surface and groundwater samples ... 43

Figure 20: Lateral distribution of major ions at the study area ... 44

Figure 21: Trend analysis of Cr(VI) at the study area prior to and after installation of the liner ... 45

(9)

Figure 24: Trend analysis of Cr(VI) at the study area prior to and after installation of the

liner ... 45

Figure 38: Trend analysis of Cr(VI) at the study area ... 49

Figure 39: Trend analysis of SO4 at the study area ... 50

Figure 40: Trend analysis of Cl at the study area ... 51

Figure 41: Trend analysis of pH at the study area ... 52

Figure 42: Trend analysis of EC at the study area ... 53

Figure 43: Calibration map ... 54

Figure 44: Mean residual head for input parameters (order of magnitude down) ... 56

Figure 45: Mean residual head for input parameters (order of magnitude upward) ... 56

Figure 46: Results of the stability of the CrO4 transition in terms of chromate concentrations over time ... 57

(10)

Figure 47: Plume movement without sorption and during active abstraction from 1 year to

27 years ... 57

Figure 48: Plume movement with sorption and during active abstraction from 1 year to 28 years ... 58

Figure 49: Capture zone due to pumping ... 62

LIST OF TABLES

Page TABLE 1: PUMP TESTED BOREHOLES ... 16

TABLE 2: LONG TERM ABSTRACTION RATES ... 16

TABLE 3: THE ANALYSED PARAMETERS AND THE METHODS ... 25

TABLE 4: INPUT PARAMETERS TO THE NUMERICAL MODEL ... 32

TABLE 5: CALIBRATION PARAMETERS ... 36

TABLE 6: VALUES OF FREUNDLICH CONSTANTS FOR THE ADSORPTION OF CR(VI) AT PHASE CONTACT TIME 72 H (WÓJCIK & HUBICKI, 2003). ... 37

TABLE 7: SOURCES OF DISSOLVED IONS AT THE STUDY AREA (BERKOWITZ, ET AL., 2007) ... 38

TABLE 8: EFFECTIVENESS OF ELECTROCHEMICAL METHOD ... 48

TABLE 9: ADJUSTED PARAMETERS TO UP AND DOWN ORDER OF MAGNITUDE .... 55

TABLE 10: MEAN RESIDUAL HEADS ... 55

TABLE 11: MASS OF CR(VI) OVER TIME WITHOUT RETARDATION ... 58

(11)

CHAPTER 1

1. INTRODUCTION

South Africa is a large-scale ferrochrome producer. The ferrochrome industry is driven by the economic growth of the country and the fluctuating prices of electricity. Although South Africa is facing challenges of economic growth, shortages of electricity and rising electricity prices, it still remains the biggest role player in ferrochrome production (Beukes, et al., 2010). When considering the global amount of mineable chrome ore, studies indicate that South Africa hosts approximately 75% to 80% of economic chromite ore reserves (Beukes, et al., 2010). Therefore, it is expected that large volumes of sludge containing Cr(VI) will be generated.

During the processes of ferrochrome production, oxygen is introduced in the process leading to the oxidation of Cr(III) to Cr(VI). Chromium in the natural environment is commonly present in two states, which are Cr(VI) and Cr(III) (Beukes et al., 2010 ). The trivalent Cr predominates and very limited hexavalent Cr is normally present. Cr(VI) is regarded as carcinogenic and can have irreversible human health effects. Cr(III) is regarded as non-carcinogenic and does not cause human health effects. It is imperative to reduce Cr(VI) in water to the targeted levels, which are dependent on the characteristics of the site, in order to reduce human health risk to acceptable levels. Although the site setting plays a major role in the target levels required, studies indicate that uncontaminated water should have concentrations of Cr(VI) less than 0.5 mg/ℓ and when the concentrations are above 0.5 mg/ℓ remedial options should be considered (DWAF, 1996) to reduce risk to groundwater users and the environment.

Several methods used to treat Cr(VI) contaminated water exist and some methods are deemed infeasible in comparison to other methods depending on site-specific conditions. Methods which can be considered to treat Cr(VI) contaminated water or prevent further leaching of Cr(VI) include in-situ biological reduction, in-situ chemical or sorption precipitation, electrochemical reduction and physical methods such as installing the impermeable liner (Fang, et al., 2012). Hazardous by-products in large amounts are produced when some of the methods to treat Cr(VI) contaminated water by means of dosing with chemical products or physical redox manipulation are used. The treatment of Cr(VI) contaminated water by means of a reducing agent (FeSO4 or FeCl2) increases

concentrations of other by-products such as SO42-, NO3- and salinity in the treated water.

These by-products are often hazardous to human health, the environment and in the production processes when not treated. Cr(VI) contaminated water and the chemical dosing thereof, should be monitored and assessed in order to avoid the possibility of causing

(12)

contamination in groundwater and the receiving environment (U.S. Environmental Protection Agency, 2000).

2. PROBLEM STATEMENT

At the study area, the groundwater contamination by Cr(VI) was initially detected in 1997, about a year after the leaking slimes dam was commissioned. Since then measures were implemented to prevent further chromium contamination. The leaking slimes dam was decommissioned in 2000 and covered with polyethylene thermoplastic to prevent leaching by rain water and Cr oxidation. All furnaces used for metallurgical processing of the chromite ore were converted to closed furnaces. Based on site-specific hydrogeological risk assessments and monitoring events, a remedial action plan was developed. Site characterisation and risk management studies lead to the conclusion that remediation was necessary. The restoration of groundwater quality to drinking water standards is generally considered impractical and technically not feasible. Corrective action following groundwater resource contamination is based on a risk-based approach. Thus, remediation is required when the contamination poses a risk to the receiving environment and human receptors. The pump and treat system was designed and implemented in February 2015. The system focuses on pumping the water from the aquifer to the surface, after which a dosing system using FeSO4 as a reducing agent is used to treat the contaminated water.

The dosing system managed to treat Cr(VI) contaminated water. However, in the process of treating Cr(VI) contaminated water, excessive amounts of hazardous by-products (SO42- and

salinity) were prominent. In order to address both Cr(VI) contaminated water and hazardous by-products at the study area, the pump and treat (P&T) system remediation plan was reviewed. Upon review, it was deemed feasible to fit another system into the existing dosing system infrastructure, in order minimize both Cr(VI) and hazardous by-products in the contaminated water. Although the remediation systems can be implemented and managed to treat the contaminated water, the effectiveness of these systems is dependent on the integrity of the polyethylene thermoplastic liner, and as a result it is essential to determine whether or not the liner is effective to prevent recharge into the slimes dam. The research presented in this dissertation seeks to examine the efficiency of the liner and the existing remediation methods.

(13)

 Determine the effect of the liner on the seepage quality observed in the monitoring boreholes.

 Assess the water quality after implementation of the remediation systems.

 Determine the extent of the Cr(VI) contamination plume in the groundwater on-site taking into account the plume mass and sorption.

 Determine if the capture zone of the P&T boreholes can capture that plume so as not to reach the impacted monitoring boreholes or receptor boreholes.

(14)

CHAPTER 2

2. LITERATURE REVIEW

2.1 Generation of Cr(VI) in Ferrochrome (FeCr) Industry

Chromium can, be found in different oxidation states from zero to six (Kamaludeen, et al., 2003). Chromium can occur naturally or as a soluble or insoluble compound. The oxidation states of chromium are named as Cr(0), Cr(III) and Cr(VI) (Beukes, et al., 2010),

Cr(0) is not naturally occurring, this metal is generated during industrial processes. Cr(III) is naturally occurring, this state of chromium is stable, immobile and has no harmful impact on human health and the environment. Hexavalent Cr exists as an oxyanion chromate (CrO42-₎

and originates from anthropogenic activities (Kamaludeen, et al., 2003). Hexavalent chromium is unstable, mobile and harmful to human health and the environment. This oxyanion is produced from activities such as ferrochrome production processes, chromium plating, textile manufacturing, leather tanning, pigment manufacturing, wood preserving and chromium waste disposal (Beukes et al, 2010).

Understanding the Cr(VI) related aspects within the ferrochrome industry requires a broad review of the production and processes undertaken. A generic process employed by the ferrochromium industry is depicted below (Figure 1).

(15)

Figure 1: A flow diagram from Riekkola-Vanhanen indicating the most common process steps utilised for FeCr production in SA (Beukes et al, 2010)

Chromite ore is converted to chromite mineral after being mined and crushed (Mining Watch Canada, 2012). This process requires high temperature submerged arc furnaces or direct current arc furnaces and oxygen can be limited or not limited from these processes. During the process a small amount of Cr(VI) can be generated.

2.1.1 Slimes dam cover

Section 24(7) of the National Environmental Management Act, 1998 (Act No. 107 of 1998), clearly states that the mining companies are liable for the management and control of hazardous waste during and after the mine closure. The impacts must be communicated, investigations to assess the extent of the impact should be implemented, action to redress the impacted environment, to achieve practical natural state or to attain acceptable sustainable development should be taken.

Section 24(7) together with environmental management plans or authorised environmental programmes, will assist in safeguarding human health and environment. For this reason, practical environmental programmes have been developed. This includes but is not limited to the installation of liners to prevent acid rock drainage through infiltration of acid seepage.

(16)

Limiting oxygen and water ingress into the slimes dam helps to prevent or manage the infiltration of acid water into the aquifer (Fourie & Tibbett, 2009). Studies by MiMi (2004) show that water covers are the best available method to minimize oxygen and water recharge into the slimes dam. This method is said to be cost effective, low maintenance and suffocates the generation of dust.

There are a number of slimes dam liners which are available and considered to be effective in the industry. One of the many liners which are cost effective, is known to be a polyethylene thermoplastic liner. The liner acts as a temporary measure to prevent water ingress and dust generation which can assist in avoiding the Cr oxidation (USEA, 1994).

2.2 Trend Analyses of Groundwater Chemistry

The purpose of the Mann-Kendall (MK) test (Kendall, 1975) is to statistically assess if there is a monotonic upward or downward trend of the variable of interest over time. A monotonic upward (downward) trend means that the variable consistently increases (decreases) over time, but the trend may or may not be linear. The MK test can be used in place of a parametric linear regression analysis, which can be used to test if the slope of the estimated linear regression line is different from zero. The regression analysis requires that the residuals from the fitted regression line be normally distributed; an assumption not required by the MK test, that is, the MK test is non-parametric (distribution-free) (Pohlert, 2018).

Studies (Kendall, 1975) further show that this software is associated with the assumption that, when there is no evident trend, the collected data is not serially corrected over time.

The GSI Mann-Kendall Toolkit is an easy-to-use spreadsheet system for analysing time series groundwater monitoring data to quantitatively determine if the measured concentrations of a chemical are increasing, decreasing, or stable over time, based upon use of the Mann-Kendall statistical method. The software can be applied to data from monitoring points for which groundwater sampling and testing have been conducted at multiple episodes over time (i.e. time-series sampling) to evaluate the concentration trend of each chemical at each monitoring location.

2.3 Transport of Cr(VI) in the Environment

It is important to understand the distribution and transportation of metals in the subsurface when dealing with toxic substances which are harmful to humans and to the environment. This involves the understanding of the mechanisms which drive the distribution and transport

(17)

(Shriver et al., 1994) in their study, illustrate how Cr(III) can be favoured and be stable in the environment in relation to Cr(VI). Cr(III) is regarded stable in water with negative Eh and the oxidation state is more favourable, and when Eh is positive the reduction conditions are more prominent. The oxidation states in relation to the potential energy are shown below (Figure 2). Another condition which determines whether the chromium will be in the state of being mobile and harmful is the pH of water. Cr(III) and Cr(VI) species are stable in basic and acidic conditions, however, this is also dependent on the Eh. Looking at Figure 3, it can be seen that a relationship exists between the Eh and the pH and different species of chromium can be identified. When Eh is positive and pH is basic the Cr(VI) is prevalent and when the reverse relationship is considered, the Cr(III) is prevalent.

(18)

Figure 3: The relationship between Eh and pH

The contaminants tend to follow the groundwater flow direction, however, some contaminants tend not to follow the groundwater flow direction and that depends on their physical, chemical and biological properties. The contaminants and groundwater follow the topography from recharge zone to discharge zone.

Groundwater and contaminants can move rapidly through fractures in rocks. Fractured rock presents a unique problem in locating and controlling contaminants because the fractures do not follow the contours of the land surface or the hydraulic gradient. Contaminants can also move into the groundwater system through macropore root systems, animal burrows, abandoned wells, and other systems of holes and cracks that supply pathways for contaminants (Allwright, 2014).

Movement of solutes which progressively detach from the source area in the subsurface is strongly dependent on the positive gradient, the physical properties of an aquifer as well as the chemical properties of the contaminant of concern. The solutes steadily migrate away from the source area and are dependent on the advection and dispersion processes via groundwater (Verral, et al., 2008). The movement of solutes in the material is conceptualised by (Strassberg, et al., 2011) (see Figure 4).

(19)

Figure 4: Three dimensional model illustrating contaminant transport by advection (Strassberg, et al., 2011)

Dispersion

Dispersion refers to a mass transferred from one point to another and is subjected to properties of the media and chemical features of the solute. Dispersion is classified as mechanical (Figure 5) and hydrodynamic dispersion (Figure 6); the two are defined by macroscopic transport parameters. Mechanical dispersion is controlled by media intrinsic permeability, substance environment dependence such as distribution coefficients in numerical models and substance dependant such as decay constant (Kalmaz & Barbierri, 1980). Hydrodynamic dispersion is a function of porous material, i.e. the ability of the fluid to pass through the material. Coefficient mechanical dispersion can also not be ruled out, and lastly, the coefficient of molecular diffusion of that medium is required (Craig, 2004). Understanding of these processes can be crucial in viewing or determining the movement of solutes from the source to receptor points. This also contributes in development of the transport models.

(20)

Figure 5: Mechanical dispersion transport on the microscopic scale (Adapted from Amy Allwright, 2014)

Figure 6: Hydrodynamic dispersion transport on the microscopic scale (Amy Allwright, 2014)

When solutes move in the soil media, there are several processes that play a major role for the solute to be transported over a short or long distance. This involves sorption, which is divided into absorption and adsorption. Absorption is the incorporation of a substance in one state into another of a different state. Adsorption is the physical adherence of ions and molecules onto the surface of another molecule. This process is the main one which can assist in determining the actual solute velocity in groundwater (USEPA, 1999).

(21)

The overall correction factor takes into account the bulk density of the media, porosity and distribution coefficient, which cannot be done using a flow model. This can be done by means of leachate testing or through literature (USEPA, 1999).

The distribution coefficient used to calculate the retardation factor for turf at the study area was based on the adsorption isotherm (Mikołajków, 2003). Freundlich isotherm shows that, when the solute concentration is introduced on the solid surface area, the solutes cover the whole surface area, until a log curve is noticed, which is subsequently followed by a straight line. This means the solutes have covered the whole surface area, however in the later stages the solute is continually introduced to the surface area and steady state is not reached (Mikołajków, 2003). This isotherm is best suited to calculate small data sets compared to Langmuir isotherm which considers large data sets and indicates that the steady state is reached when the whole surface area is covered by solutes. Both isotherms can be used to represent the early stages of the solutes being introduced onto the solid. However, Langmuir isotherm is best suited for early and late concentration rates.

The distribution coefficient used to calculate the retardation factor is known as 1st_sorption

constant in the transport model and was sourced from the experiments conducted by (Wójcik & Hubicki, 2003). The experiment showed a value of 20 for the distribution coefficient (Kd) of the Xeolith material in alkaline conditions by means of using Freundlich isotherm.

2.4 Numerical Flow Models

A numerical model is a mathematical representation of a real system in a simplified form and takes its physical properties into account to make calculations and inferences about system behaviour. Physical properties are relevant in a flow model while chemical properties are relevant in a geochemical model (Reilly, et al., 2004). It is realised that models cannot address all problems, hence it is important to evaluate the model for its use and consider the assumptions related to the model.

Evaluation of the model is fundamental. Realistic and conclusive decisions can be made with respect to both water quality and water quantity pertaining to the future response of the groundwater system. The evaluation process provides an indication of system behaviour for management of the system. Part of the evaluation process is the understanding of the application of the model. There are several potential applications of the model some of which are listed below (Milovan & Randall, 1992):

 Design and/or evaluation of pump-and-treat systems  Design and/or evaluation of hydraulic containment systems  Evaluation of physical containment systems (e.g. slurry walls)

(22)

 Analysis of "no action" alternatives

 Evaluation of past migration patterns of contaminants  Assessment of attenuation/transformation processes

 Evaluation of the impacts of non-aqueous phase liquids (NAPL) on remediation activities (dissolution studies)

 Determining the effectiveness of the slimes dam liners

Given that abstraction of groundwater is taking place at the study area, radial directions are expected. This means that there are limitations to the model that can be used: simple models such as one dimensional and two dimensional models can be deemed not suitable for modelling the study area, given that they focus on simple calculations to assess the groundwater behaviour (one-dimensional model) and assessing only the vertical flow processes (two-dimensional). A three dimensional model focuses on complex calculations and considers both vertical and horizontal flow processes and it is deemed suitable to simulate the groundwater behaviour at the study area (Allwright, 2014).

When working with a model, it is important to note that for one to acquire defensible results, assumptions are considered to play a major role. Assumptions which are considered in most models are the boundaries, hydraulic heads and hydrological capabilities such as wells, particle tracking, recharge and groundwater evaporation (Reilly, et al., 2004). The numerical model at the study area will be used to forecast the future and understanding of groundwater systems and how the pumping at the study area controls the flow paths.

A number of mechanisms are considered when a transport model is developed. Complex models are developed as a mathematical representation of the hydrogeological system to assess advective, diffusive, advective-dispersive, density-dependent, multiphase and other transport problems. Transport modelling is similar to groundwater flow modelling, in that all the same steps are required, i.e. defining model objectives and complexity, data collection, developing the conceptual model, developing the numerical model, refining the numerical model, and applying the numerical model. Transport modelling is dependent on a groundwater flow model as simulation of contaminant transport requires the simulated flow field calculated by a groundwater flow model. A transport model is built upon the basis of a groundwater flow model (Amy Allwright, 2014).

(23)

CHAPTER 3

3. SITE SETTING

3.1 Site Location

The study area is situated in North West Province, South Africa. The study area is surrounded by industrial, agricultural and informal land uses. The hydrocensus study indicates that groundwater users are present within a 2 km radius of the site. Industrial land use is limited to the east and north of the contamination source. The site locality study area layout can be seen in Figure 7.

Figure 7: Site location

Chromium (VI) contaminated groundwater at the study area was reported in 1997. The contamination followed after the Old baghouse slimes dam liner ruptured and Cr(VI) leachate infiltrated into the groundwater. The initial detection of Cr(VI) in 1997 was reported following a hydrocensus study conducted at the site in 1997. The leaking baghouse slimes dam was decommissioned in the year 2000 and covered with polyethylene thermoplastic to prevent leaching by rain water and Cr oxidation. All furnaces used for metallurgical processing of the chromite ore were converted to closed furnaces.

(24)

Considering the Cr(VI) contamination levels found during hydrocensus at the study area, a risk management approach indicated that a remedial action plan was necessary for the study area. The risk assessment evaluates practices with traditional site investigation and remediation selection activities. However, cost-effective measures for the protection of human health and environmental resources are important. These risk-based corrective actions can address any of the steps in the exposure process, such as removing or treating the residual source, interrupting contaminant transport mechanisms or controlling activities at the point of exposure.

Based on the site classification for contaminated land within a framework of the South African National Environmental Waste Act (Act 59 of 2008) the Baghouse Slimes was classified as a high risk site.

Prior to Implementation of Two (2) Remediation Methods

After the study area was deemed to be hosting hazardous facilities, in view of waste control and environmental impact management, a remediation strategy was developed. The formulated strategy began to address the handling of waste and sought to characterise and remediate the contamination at the study area to the acceptable levels. This began by reviewing the historical data and installing abstraction groundwater boreholes to the existing groundwater monitoring system.

During the review of the existing monitoring groundwater system, it was determined that an additional four groundwater monitoring boreholes to the existing thirteen (13) monitoring boreholes should be installed to act as abstraction groundwater boreholes, while the existing thirteen boreholes will act as source, plume and impact monitoring boreholes. Beside the seventeen existing monitoring boreholes, two monitoring boreholes (PIM5 and IM1) were found at the study area and were only monitored and sampled in 2015.

The thickness of the vadose zone and geology was determined from the geological logs of the four drilled monitoring boreholes. The overview of the soil profile and geology descriptions, vadose zone and geology thicknesses are presented below (Figure 8).

(25)

Figure 8: Overview of the vadose zone and geology of the area

A total of four abstraction boreholes were installed in order to form part of the existing groundwater monitoring and treatment system. The boreholes were subjected to pumping tests to attain suitable long term abstraction rates, to allow maximum drawdown. This in turn was required to intercept and retrieve the maximum allowable volume of Cr(VI) contaminated groundwater. The boreholes subjected to the aquifer test are shown in Table 1 below. A summary of the FC results is shown in Table 2.

Based on the drawdown data below, boreholes BH-BH1, BH-PM2 and BH-PM3 were selected to be suitable to be long term abstraction boreholes. Borehole PMA was not

(26)

recommended for long term abstraction, as it will run dry within minutes even at very low pumping rates and is therefore not suitable for long term abstraction.

Aquifer Test Summary

Three (3) low flow pumps were installed in BH-BH1, BH-PM2 and BH-PM3. The pumps were determined to have a flow rate of less than 0.5 ℓ/s. As a safety measure, the pumps were set to automatically switch off when the water level reaches the pump inlet to prevent pump burnout. The volume of water permitted to be abstracted ranged between 43 m3_{/day and}

67 m3_{/day. The total amount abstracted is 130 m}3_/day.

Table 1: Pump tested boreholes

Borehole Number Borehole Depth (m) Water Strikes (mbgl) Pumping Time (min) Constant Rate Discharge (ℓ/s) Recovery Time (min) Static Water Level (mbgl) Drawdown (m) BH-PM2 30 21 210 0.25 30 13.51 0.14 BH-BH1 30 21 180 0.25 40 12.04 0.27 BH-SM3 58.5 21 35 0.25 90 12.56 12.09 BH-PM3 40.5 21 120 0.25 60 14.24 0.1

Table 2: Long term abstraction rates

Borehole Number Transmissivity (m2_/d) Sustainable Yield (ℓ/s) Sustainable Yield (m3_/day) Sustainable Yield (m3_/week) Sustainable Yield (m3_/month) BH-PM2 130.0 0.2 17.3 121.0 3628.8 BH-BH1 62.9 0.5 43.2 302.4 9072.0 BH-PM3 79.1 0.8 67.4 471.7 14152.3 Total 1.5 127.9 895.1 26853.1

Based on the abstraction volumes calculated, a Pump and Treat System, which constituted three abstraction groundwater boreholes and two settling ponds with a dosing tank was developed. The system was implemented in February 2015. The pump and treat system

(27)

 Two settling ponds (pre-treatment and post-treatment).  One dosing pump.

 Two tanks storing and transmitting aqueous FeSO4 to the dosing pump.

Prior to implementation of the system a geochemical model to assist in calculating the volume of product to be applied to allow the reduction of Cr(VI) to Cr(III) by means of dosing with FeSO4 or FeCl2 was developed. The following parameters were used:

 Volume of Cr (VI) contaminated groundwater of 130 m3_/day;

 Concentration of Cr (VI) contaminated groundwater containing 5 mg/ℓ;  Average hydrochemistry of extracted groundwater; and

 TDS of Cr (VI) contaminated groundwater of 2600 mg/ℓ

Based on the geochemical modelling results, a volume of 0.13 m3_{/day (FeCl}

2) solution was

calculated be added to a volume 100 m3_{/day of Cr(VI) contaminated groundwater on a daily}

basis to allow for reduction of Cr(VI) to Cr(III). Additional to this, low grade chromite ore and scrap metal were added to settling ponds in the pilot system to further promote the precipitation of Cr(VI) to a Cr(III) oxide form.

Associated risks with ferrous chemicals

Although highly effective in reducing Cr(VI) to Cr(III), the use of ferrous chemicals has numerous disadvantages. These include:

 Their use increases the total dissolved solids (TDS) content of the process and waste water. Fe(II) is removed by oxidation to Fe(III), which consequently forms an Fe(III) hydroxide. This hydroxide precipitates from solution at the pH levels relevant to the FeCr process and waste water. However, the chloride or sulphate remains in solution, causing the increase in TDS.

 The abovementioned increase in TDS could result in increased scale build-up in pipes, spray nozzles of wet scrubber systems, and other equipment.

Although the ‘major’ environmental and health risk, i.e. Cr(VI), is effectively dealt with during the reduction of Cr(VI) to Cr(III) by Fe(II), the increased TDS and chloride or sulphate load could result in increased salination of surface and groundwater, due to potential process and waste water leakages. Although salination of surface and groundwater is not regarded as crucial as Cr(VI) contamination, it is certainly not acceptable.

P&T system for the current study area indicated that monitoring of the system has shown performance in line with the remediation expectations. Residual concentrations of Cr(VI)

(28)

were found to be below detectable levels. However, the hazardous by-products were deemed to be of concern. Hence, an enhancement of the remediation system was considered and concluded to be required.

An electrochemical method was selected at the study area in order to assist the already existing chemical Fe(II) reduction method. This method was selected given that the chemical reduction method had disadvantages in the process of treating the Cr(VI) contaminated water. Although the electrochemical method is currently employed by South African FeCr producers, there is not sufficient research that indicates the effectiveness and environmental soundness of the method. Hence the traditional chemical Fe(II) reduction is currently a preferred method (Beukes, et al., 2012).

Electrochemical method is an alternative method which was considered to be competitive and effective in comparison to chemical reduction method. Studies by Wittbrodt & Palmer, 1992 show that residual Cr(VI) is difficult to remove as the concentration decreases during the pump and treat remediation process. Moreover, studies by Gonzalez, et al., 2003 further show that agents used for reduction of Cr(VI) generate toxic sludge that may be a challenge to remediate.

The electrochemical research (Barrera-Diaz, et al., 2011) indicates that the Cr(VI) is reduced to Cr(III) without introducing chemicals in the process of treating Cr(VI) contaminated water. Subsequently chromium hydroxide is formed and is attracted to the electrode surface area.

Based on the study by (Fang, et al., 2012), it can be deduced that the electrochemical reduction method is best suited to lower Cr(VI) concentrations without generating hazardous by-products as is the case with the chemical reduction method. Electrochemical effectiveness was investigated in detail (Fang, et al., 2012) and it was found that the removal of Cr(VI) is driven by various properties such as the type of cathode material, solution pH, PANI film thickness and electrolyte temperature on the Cr(VI). When the solution pH is low and the PANI film thickness is suitable, the potential to remove the Cr(VI) in the solution is increased. Although the method can be regarded as beneficial in removing the Cr(VI) and no hazardous by-products are generated, it should be noted that the method can pose operational glitches.

In August 2016, an electrochemical unit was installed at the study area. This remediation method was aimed at reducing both the Cr(VI) contaminated water, high dissolved ions,

(29)

implementation of electrolysis prior to treatment and in situ electrolysis at boreholes, and the electrochemical system are in its pilot phase and will be up-scaled if considered successful.

In the case of the electrochemical system, the geometry, material and reactor configuration play a major role in electro-reduction of Cr(VI). For the purpose of this study, only the electrode material was considered. Barrera-Diaz, et al., 2011, in their research, have illustrated that aluminium and stainless steel do not provide less time of reduction of Cr(VI) compared to iron and copper, where approximately 16% less time is required to reduce Cr(VI) to less than 1 mg/ℓ. The use of iron electrode was proven to be successful in comparison to the copper electrodes. The copper electrodes experience the limitation of mass transfer, while the iron electrode has been proven to adequately handle the mass transfer. Hence the iron electrode was selected to be suitable to build the electrochemical system implemented at the study area. The iron electrodes have the ability to liberate iron ions into the Cr(VI) contaminated water and in turn the ions act as the removal agents of Cr(VI) in the solution. The process of chemical reduction which is followed is represented in the following equations (Barrera-Diaz, et al., 2011).

Anode

Fe(s) Fe2+(aq) + 2e

-Cathode 2H2+

(aq) + 2e- H2(g)

2 H20(aq)+ 2e- H2(g) + OH-(aq)

Bulk solution 7H+_{+ 3 Fe}2+

(aq) + HCrO- 3 Fe2+(aq) + Cr3+(aq) + 4H20(l)

3.2 Climate and Drainage

The study area is located in the Highveld climatic zone where summer rain occurs mainly in the form of thunderstorms with the majority of rainfall events occurring between January and April. Winter is cool and dry. Evaporation volumes in the area on a monthly basis greatly exceed rainfall volumes. Temperatures in the region tend to be warm to mild. The average maximum temperature is 25°_{C and the average minimum temperature is 9}°_{C. The average}

(30)

Figure 9: Mean monthly rainfall and evaporation data as recorded at the Hartebeespoort Dam Meteorological Station (A2E001) over a 17-year period (Department of Water and

Sanitation, 2017)

3.3 Geology

The well-known largest intrusion on earth is known as Bushveld Complex, with lateral extent over 65 000 km2_{and up to 8 km thickness (Von Gruenewaldt, 1977). The intrusion is}

confined within the boundaries of the Transvaal sedimentary basin. The intrusion together with the sedimentary rock formation is due to intracratonic rifting (Eriksson et al. 1991). About 80% of world’s resources of PGE and the majority of resources in Cr and V is embedded in the Bushveld Complex (Naldrett, 2004).

Studies by Kruger (2005) indicate that the Bushveld Complex includes four distinct igneous suites. These suites are known as Uitkont Complex, Rooiberg Felsites, Rustenburg Layered Suite and the younger Lebowa Granite Suite (Naldrett et al., 2009; 2012). The Bushveld Complex is divided into three limbs, western, eastern and northern limbs (Wilson and Chunnett, 2006; McDonald et al., 2009; Maier et al., 2013).

The western and eastern limb of Bushveld Complex comprises abundant chromite, as opposed to its northern limb. The study area is situated within the western limb of the Bushveld Igneous Complex. The western limb extends for approximately 200 km along an arc from near Thabazimbi to near Pretoria. The mineral deposit on-site is that of the middle group (MG) chromite seams. The Chrome Middle Group (MG0, MG1 and MG2) seams are currently being exploited by the approved mining operations on-site. The seams have an

(31)

The Bushveld Complex consists of basal mafic and ultramafic rocks, named as the Rustenburg Layered Suite and divided into the Marginal zone, Lower zone, the Critical zone (which consists of chromite deposits and the Platinum Group Metals), Main zone, and Upper zone. The economic minerals such as Chromium and Vanadium are contained in the Merensky Reef, which is located within the Rustenburg Layered suite of the Bushveld Complex (Wilson & Anhauesusser, 1998).

The local geology is typical of the Bushveld complex based on the 2526 Rustenburg 1:250 000 geology series map (Figure 10). The area is predominantly underlain by the gabbro-norite of the Rustenburg Layered Suite of the Bushveld Igneous Complex with sporadic areas underlain by diabase. The gabbro-norite is stratigraphically located in the Main zone of the Rustenburg Layered Suite of the Bushveld Igneous Complex. The Main zone is a succession of norite and gabbro-norite with minor anorthosite and pyroxenite. The area is characterised by thick black clay turf.

(32)

(33)

3.4 Hydrogeology

The study area is underlain by a weathered and fractured aquifer. This is expected given that the Bushveld Igneous Complex comprises a shallow weathered aquifer with deeper semi-confined fractured bedrock. The igneous rock is classified as crystalline and is associated with fractures and lineaments. Groundwater at the study area occurs in the fractures network (Wanner, et al., 2011). The underlying aquifers in the area are semi-confined aquifers in the weathered zone of pyroxenite, norite and anorthosite. Based on geological logs, the study area is underlain by pyroxenite, norite and anorthosite. The dykes run across the study area and three major faults run parallel to each other (see Figure 11). These dykes and faults act as the conduit for groundwater flow (Bense, et al., 2013) and have an effect on water levels. The abstraction of groundwater for the purposes of groundwater remediation and mining activities also influence the groundwater flow direction. The groundwater flow direction was calculated to be in a north-westerly direction as shown in Figure 11below.

(34)

CHAPTER 4

4. METHODOLOGY

4.1 Monitoring and Sampling of Groundwater Boreholes and Settling Ponds

Groundwater monitoring and sampling was effective from 1995 prior to the detection of the Cr(VI) contamination in 1997. The surface water sampling of the settling ponds started in 2015 after the remediation systems were installed at the study area. The sampling procedure is outlined below.

Sampling procedure

Groundwater samples were collected by purging (pumping and bailing where possible). The groundwater level was measured using a dip meter before introducing any equipment into the borehole. The groundwater samples were collected by means of a bailer/pumping to draw water from the borehole. The water samples were collected in a 1.5 litre pre-cleaned plastic bottle and a 100 millilitre acid prepared plastic bottle. All samples were kept on ice during fieldwork sampling and later in a refrigerator until delivered to the laboratory. All relevant field information collected was logged. Chain-of-custody documents were prepared as part of the QA/QC and submitted to the laboratory to allow tracking of the samples through the process. The groundwater samples were submitted for major cation/anion analyses, including some constituents associated with activities taking place on the study area.

Sampling of the settling ponds was conducted on a monthly basis to assess the P&T and electro-reduction remediation systems. The abstraction boreholes were selected to pump Cr(VI) contaminated water into the built settling ponds for treatment purposes. These boreholes were drilled within the plume. The source, plume and impact monitoring boreholes were selected to monitor groundwater, in order to assess the groundwater status at certain areas in such a way that an overview of the pollution status of the area could be obtained.

Source monitoring boreholes were selected or placed close to, or in, the source of contamination to indicate its impact on the groundwater chemistry in its immediate vicinity. The plume monitoring boreholes were selected or placed in the primary groundwater plume’s migration path to evaluate the migration rates and chemical changes along its

(35)

Figure 12: Site showing monitoring points in relation to the Old baghouse slimes dam

4.2 Trend Analyses

Water samples were collected from the existing monitoring boreholes, from the surface water body (dam) and from the settling ponds (A and B). All collected water samples were submitted to a SANAS accredited laboratory. The method used to analyse the cations was the inductively coupled plasma optical emission spectrometry, while the anions were analysed by spectrophotometry. Cr(VI) was analysed using ion chromatography.

The trend analyses were conducted in order to determine whether the installed liner is capable of reducing the generation of Cr(VI) and minimizing further Cr(VI) leachate. The trend analyses for the study area were conducted prior to and after installation of the liner. The analysed parameters and the methods used for analyses are presented in Table 3 below. To determine the upward and downward trend, the Mann-Kendall (MK) test was used.

Table 3: The analysed parameters and the methods

Analyses in mg/ℓ Method

(36)

pH – Value at 25°C WLAB001

4500-H+ pH Value: Standard Methods for Examination of Water and

Wastewater; 20th Edition, Instrument manual

Electrical Conductivity in mS/m at

25°C WLAB002

2510 Conductivity: Standard Methods for Examination of Water and Wastewater;

20th Edition, Instrument manual

Chloride as Cl WLAB046 Instrument manual

Sulphate as SO4 WLAB046 Instrument manual

Hexavalent Chromium as Cr6+ _WLAB032

3500-Cr Chromium; Colorimetric Method: Standard Methods for Examination of Water and Wastewater;

20th Edition, Instrument manual

Iron as Fe WLAB015

3500-Fe Iron; Phenanthroline Method: Standard Methods for Examination of Water and Wastewater; 20th Edition,

Instrument manual

4.3 Conceptual Site Model

The conceptual site model was considered in terms of the source, pathway and receptor methodology. The source of Cr(VI) contamination at the study site is associated with the decommissioned baghouse slimes dam. Since the baghouse slimes dam was decommissioned in the year 2000 it has been covered with polyethylene thermoplastic to prevent leaching by rain water and Cr oxidation. Based on the literature review performed and data obtained from the study area, a conceptual site model was developed to provide a conceptual understanding of the hydrogeological and geological characteristics of the aquifer system.

Residual Cr(VI) contamination in the sub-surface acts as a secondary source of contamination. The Cr(VI) plume has migrated from the study area to the impacted supply groundwater boreholes BH-PIM1 and BH-PIM2 through the weathered zone and fractures underlying the site. The weathered zone and fractures underlying the study area are seen to be the preferential pathway and the impacted supply groundwater borehole is a receptor of concern.

(37)

were calculated for the regolith. The hydraulic properties and geological layers underlying the study area are presented in Figure 13. The source of contamination was identified to be a decommissioned slimes dam which was afterwards covered with an impermeable liner. The contamination leaching into the groundwater is regarded as the secondary source. The contamination migrates via the preferential pathways such as the faults and fractures into groundwater. Based on the measured Cr(VI) contamination in the impact monitoring boreholes, it can be deduced that a linkage exists between the source, pathway and receptor. The concentration levels of Cr(VI) presented in the conceptual model in Figure 13, indicate Cr(VI) concentrations before the polyethylene thermoplastic liner was used to cover the slimes dam. The concentrations in the conceptual model in Figure 14, indicate Cr(VI) after the polyethylene thermoplastic liner was used to cover the slimes dam. The presence of the underlying clay liner could not be confirmed.

(38)

(39)

(40)

4.4 Numerical Modelling

A conceptual model was used to develop the numerical model which was used to assess groundwater flow, mass of the plume and the capture zone of the plume. The finite difference model was used to simulate the conditions at the study area. This was done by means of using the Groundwater Modelling System (GMS10.0).

4.4.1 Model objectives

A local groundwater flow model was developed based on the available aquifer parameters and model assumptions to calculate the potential hydraulic trends of the groundwater system in relation to the contaminant of concern. The numerical model was developed in order to determine the groundwater flow direction, the mass of the plume and the capture zone of the abstraction system.

4.4.2 Assumptions

 The geological structures were not taken into account during modelling.

 The Dry deposition Cl used in the mass chloride programme was unknown and as a result the Dry deposition Cl from the Recharge programme was used.

 The abstraction remediation boreholes are continuously pumping without interruption.  To calibrate the model, eight (8) boreholes were used as observation boreholes to

calibrate the numerical groundwater flow model.

4.4.3 Data sources

The development of the hydrogeological conceptual and numerical groundwater model was based partly on the following information and data made available to the project team or gathered as part of the groundwater investigations.

 1:250 000 Geology Map of the area (2526 Rustenburg, Council for Geoscience).  1:500 000 Hydrogeological Map sheet (2526 Johannesburg, DWA).

 Digital elevation model (DEM) converted into a 90 m x 90 m grid.

The upper (first) metre below the surface consists of turf. This turf layer is anticipated to have a reasonably high hydraulic conductivity and a seasonal water level is expected in this layer, especially after high rainfall events. Flow in this perched aquifer is expected to follow

(41)

geological layers are vertical presented (Figure 15). The main aquifers supplying groundwater in the area are fractured aquifers, which have a lower hydraulic conductivity. The permanent groundwater level resides in this unit and ranges between 7.16 m and 20.53 metres below ground level. The groundwater flow direction in this unit is influenced by the regional topography and is generally from high lying areas to the surface drainage courses.

Below a few tens of metres the fracturing of the aquifer is less frequent due to increased pressure. This results in an aquifer of lower hydraulic conductivity and very slow groundwater flow velocities (pyroxenite or anorthosite or gabbro-norite).

Hydraulic conductivity in the constructed models was decreased by an order of magnitude in each successive layer. This was performed based on the work by (Wang et al., 2009) and (Cheema, 2015) which shows that hydraulic conductivity often decreases exponentially with depth. The model inputs are presented in Table 4 below.

Although the most relevant aquifer parameters are optimised by the calibration of the model, many parameters were calculated and/or judged by conventional means. The following fixed assumptions and input parameters were used for the numerical model of this area (Table 4).

(42)

Table 4: Input parameters to the numerical model

Model Parameter Value Unit Reason

1 Recharge to the aquifer 0.0001 m/d Calculated (5%)

2 Evaporation 0.0043 m/d Calculated (Department of Water and Sanitation, 2017)

3 Boundaries River and escarpment - Existing boundary conditions present at the site that would potentially include modelled impacts

4 Refinement 20 m Based on the scale of the baghouse slimes dam

5 Grid dimensions 90 x 90 Cell count Product of the grid refinement

6 Hydraulic conductivity Layer 1 = 0.3 Layer 2 = 0.1 Layer 3 = 0.01 m/d Du Toit, G.J. (2010) 7 Hydraulic anisotropy

(vertical) 10 - Anderson et al. (2015)

8 Effective porosity

(30% for regolith)

5 declining to 3 with depth in each layer

(43)

Model Parameter Value Unit Reason

11 Mean residual head

error 1.97 m Head error statistics

12 Head error range 10 m Calculated as 10% of the difference between the maximum and

(44)

4.4.4 Model boundaries

Model boundaries were identified on the premise that natural topographical features such as mountains and rivers served as either no-flow boundaries or constant head boundaries. In addition, model grid refinement at the decommissioned baghouse slimes dam allowed for an accurate and detailed solution of the model matrix in an area (Figure 16).

The same finite-difference flow model was used as a basis for the contaminant transport model; i.e. the 3-dimensional, regional three-layer steady-state groundwater model. However, due to the intensive computer calculations required for each grid cell and associated computer resource limitations, the model grid was adapted to be coarser away from the baghouse slimes dam site, with decreased cell sizes (more refined model grid) towards the baghouse slimes dam (90 m x 90 m).

(45)

4.4.5 Recharge

Calculation of recharge was done by means of using chloride mass balance method for the local area. This was done by means of using the parameters described below (Van Tonder & Xu, 2000) in the recharge programme. The average rainfall = 668.5 mm/a (DWS, 2017). The Cl in rain = 0.2207 mg/ℓ for inland (The Cl in rain water was unknown; as a result the Cl from the Recharge programme was used). The dry deposition Cl = 0.1 mg/ℓ x (Cl of rain) for inland if no forest exists (The Dry deposition Cl was unknown; as a result the Dry deposition Cl from the Recharge programme was used.). The least contaminated boreholes (PIM3 and IM3) located outside the plume were used to calculate the Cl in groundwater = 35 mg/ℓ (From non-impacted monitoring boreholes)

q = average ground water recharge Clr = Cl in rain

Cp = Average annual precipitation Dc = Dry deposition Cl

Cgw = average Cl in groundwater

4.4.6 Evaporation

Evaporation data obtained from the Department of Water Affairs was applied in the first layer of the model, as it is expected that the evaporation can only occur on the shallow layer (regolith). The evaporation was converted from mm/a (1576.1 mm/a) to m/d (0.0043 m/d) prior to being imputed in the model.

4.4.7 Calibration

To render a model successful or not, a model should be calibrated prior to being used to calculate the possible trends of the groundwater system. Calibration also aims to prove that the input parameters and boundary conditions are coupled to represent the aquifer system as these parameters produce the same calculated water levels as were measured in the field. The model input data can be modified to fit the observed heads and flows (Reilly, et al.,

q

(46)

2004). However, if the data does not yield the desired results, then the conceptual model can be re-evaluated. When the observed values are far off from the calculated values, it indicates that there is an error in the parameters used in the conceptual model. The best fit and the extent to which aspects of the simulation are incorporated in the model are both crucial in determining how well the model is calibrated (Reilly, et al., 2004). The parameters which can be calibrated are shown in Table 5 below.

Table 5: Calibration Parameters

Group Parameter

Hydraulic properties Hydraulic conductivity

Evapotranspiration, Recharge

Boundary conditions Precipitation

To calibrate the model, eight boreholes were used as observation boreholes to calibrate the numerical groundwater flow model. These boreholes were selected on the basis that a water levels trend was obtained from the early to latest stages of monitoring when compared to the remaining eleven monitoring boreholes. Four groundwater boreholes of the eight are pumping boreholes and as a result perfect calibration was not expected. If these values fall within the allowed calibrated error range (10% of the maximum minus the minimum calculated groundwater head elevation) the model is considered to be successfully calibrated (Reilly, et al., 2004). The elevations were retrieved from a DEM and were used to create the elevations for each cell within the model grid. Once each cell was allocated an elevation, the measured water levels in metres above mean sea level from the observation boreholes were used to calibrate the model.

4.4.8 Sensitivity analysis

The model requires to be stressed differently from the calibrated conditions. Moreover, calibration is difficult as values for hydrologic parameters, stresses and boundary conditions are typically known at only a few nodes and are associated with uncertainty. In addition, there is even uncertainty about the geometry of the hydrologic system being analysed. In order to reduce the uncertainty, it is essential to subject the (already) calibrated model to a so-called sensitivity analysis. Studies by Reilly, et al., 2004 show that the sensitivity analysis

(47)

4.5 Transport Modelling

As proven by (Zheng & Wang, 1999) MT3D transport package can be used to simulate advection and dispersion of solute, and the retardation factor of turf should be considered. This programme is designed to work with finite difference flow models. In this programme, the limitations of the model, assumptions and strengths are considered. Given that this programme considers the mentioned factors, it can be said to be the best approach to solve the mathematical problems which are representative of the conceptualised study area.

The following inputs and calculations were used to determine the retardation factor for the Cr(VI) concentrations at the study area (Table 6). A source concentration of 10% of Cr(VI) was used to develop the transport model and this is equivalent to an average of Cr(VI) concentrations of 1.0 mg/ℓ, given that the software is unable to calculate small figures. The movement of contamination was assumed to be controlled by pumping that is occurring at the study area at a pumping rate of 130 m3_{/day from the three abstraction boreholes. The}

model inputs presented (Table 6) were also used to develop a transport model.

Table 6: Values of Freundlich constants for the adsorption of Cr(VI) at phase contact time 72 h (Wójcik & Hubicki, 2003).

Parameter Value Reason

Bulk density 1.1 (Agricultural Information Bank, 2018)

Total porosity 58 (Agricultural Information Bank, 2018)

Distribution Coefficient (L3_/M) ₂₀ _{(Wójcik & Hubicki, 2003)}

1st_{sorption constant} ₂₀ _{(Wójcik & Hubicki, 2003)}

2nd_{sorption constant} _72hrs _{(Wójcik & Hubicki, 2003)}

Concentration of Cr(VI) 10% Groundwater concentrations

4.6 Mass of the plume

Modflow for flow solution M3TDS concentration and contamination dataset were used to calculate the mass of Cr(VI) plume and this was calculated with retardation and without retardation. The following formula derived from modflow was used to calculate the mass of the plume over time via transport model. The borehole logs were also considered during the calculation of the mass of the plume.

Mass Calculation

(Top – bottom of the model) x length of the plume x width of the plume x porosity x concentration of the plume x unit conversion factor

Process Enhancement of a Cr (VI) Remediation Method to Minimize the Hazardous By-products in the Treated Water